View Single Post
Old 15th March 2013, 17:51   #45
Forum King
MrSinatra's Avatar
Join Date: Dec 2004
Location: WKPS, State College
Posts: 5,612
Send a message via AIM to MrSinatra
Originally Posted by Batter Pudding View Post
Never understand why you do the huge replies.... a quick summarised reply is clearer to read. I am not trying to challenge you. This is a community working together.
i understand there is no malice in this comment, but i respond the way i think is helpful to "the community working together" and if anyone doesn't like it, they simply don't have to read it. i am at a loss as to why anyone thinks someone else should write the way they approve of, either substantively or stylistically, (outside of normal givens, like comprehensibleness). frankly, it smacks of hubris imo.

Originally Posted by Batter Pudding View Post
Yes, a single Sent To fails. As I noted above a few times. But clearly didn't make that clear enough. Sorry for my dodgy English.
if i missed it earlier, my mistake. the threads gotten long and grown quickly.

Originally Posted by Batter Pudding View Post
It is the Send To dialog that the fail appears through. If that is a list of one, or a list of 39, it fails at the same point. Just before "Done" should be appearing on the screen. And as noted by DrNo this is visible at the point it gets lost in the call stack. I assume as the Gracenote library returns from the call, it falls over. But this will be still out of sight of the Winamp guys at that point. So still in that grey area of - is it a bad function or bad data passed in?
i follow. i should correct myself, on my test file it fails as "processing" not "processed."

Originally Posted by Batter Pudding View Post
Cool. The whole baseball bat of the toolkits you have. Mad isn't it? What is "special" here? Very odd. I have put sooooo many track through the Winamp Auto-tagger over the years, and yet this little set has something in common that Winamp don't like. This is why I would be surprised if it was the Winamp code itself. It is something special that these tracks are triggering.
yes, i tend to agree. i have tried everything i can think of.

Originally Posted by Batter Pudding View Post
When I did my tests this morning the main thing I did was to let Winamp retag the file. And this action alone I would have expected the whole Tag part of the file to get re-written. And if you have removed the tags then you have taken that further. Yet still the puzzle is left in the actual MP3.
I am suspicious of the files. If they come form P2P then there can be gaps in there. Or the file can just be missing bits. I don't know an MP3 is encoded, but if there are data fields for lengths of the actual music track, maybe the music track is actually of the wrong length?
i don't think it has anything to do with tags at all, given the results. i think it is something in the files properties, and missing frames is a likely culprit, as there is a "light" kind of audio fingerprinting done by autotag. so if it is failing in "processing" it could be due to missing frames in the middle of the file from bad p2p as you suggest. but why then does it work with alt+3?

and the headers do not show an error between file length/data length, etc... sometimes i get header errors in the Xing header where it reports one size while the file is actually another, etc, but that doesn't seem to be the cause here, b/c even if it were, it still crashes after the exam/fixed.

Originally Posted by Batter Pudding View Post
I am aware that Gracenote has an algorithm that also involves "listening" to the track to identify it. So I guess that something about an attempt to "Listen" to these tracks is providing bad data to their function.

Still bonkers though
yeah, thats the audio fingerprinting. its not a true audio fingerprint, but like a "quick one." u can see this is how it works though, b/c if u take a file with no tags, and random file name / file location, sometimes it will still ID it correctly.

the formula does seem to be:

some "unique" and as yet unknown characteristic of these files, probably in the encoding, + the send to function and gracenote = a crash everytime.

are they all vbr? i wonder if they all share the same encoder? (some are crappy!)

i also wonder where else we could "send to" these files to see if other functions crash?

BUG #1 = Winamp skips short tracks
Wish #1 = Multiple Column Sorting
Wish #2 = Add TCMP/Compilation editing
MrSinatra is offline   Reply With Quote